Class prediction results using 3 miRNA biomarkers
| Phenotypic class . | Microarray . | Quantitative RT-PCR . | ||||||
|---|---|---|---|---|---|---|---|---|
| SVM, no. (%) . | RF, no. (%) . | LDA, no. (%) . | Total . | SVM, no. (%) . | RF, no. (%) . | LDA, no. (%) . | Total . | |
| ET | 24 (88.9) | 24 (88.9) | 25 (92.5) | 27 | 9 (90) | 9 (90) | 9 (90) | 10 |
| NO | 30 (100) | 25 (83.3) | 29 (96.7) | 30 | 9 (90) | 10 (100) | 10 (100) | 10 |
| Average | 54 (94.7) | 49 (86.0) | 54 (94.7) | 57 | 18 (90) | 19 (95) | 19 (95) | 20 |
| Phenotypic class . | Microarray . | Quantitative RT-PCR . | ||||||
|---|---|---|---|---|---|---|---|---|
| SVM, no. (%) . | RF, no. (%) . | LDA, no. (%) . | Total . | SVM, no. (%) . | RF, no. (%) . | LDA, no. (%) . | Total . | |
| ET | 24 (88.9) | 24 (88.9) | 25 (92.5) | 27 | 9 (90) | 9 (90) | 9 (90) | 10 |
| NO | 30 (100) | 25 (83.3) | 29 (96.7) | 30 | 9 (90) | 10 (100) | 10 (100) | 10 |
| Average | 54 (94.7) | 49 (86.0) | 54 (94.7) | 57 | 18 (90) | 19 (95) | 19 (95) | 20 |
Statistical classifiers were used to assign phenotypic class (ET or NO) based on microarray (N = 57) or quantitative RT-PCR (N = 20) expression profiles using the 3-member miRNA list (miR 10a, miR 148a, and miR 490 5p). Statistical classifiers used include SVM (Support vector machine), RF (Random Forests), and LDA (linear discriminant analysis).